Dana Pe'er (born 1971), Chair and Professor in Computational and Systems Biology Program at Sloan Kettering Institute is a researcher in computational systems biology. A Howard Hughes Medical Institute (HHMI) Investigator since 2021, she was previously a professor at Columbia Department of Biological Sciences. Pe'er's research focuses on understanding the organization, function and evolution of molecular networks, particularly how genetic variations alter the regulatory network and how these genetic variations can cause cancer.
Early life and education
Pe'er was born in Israel.[7] Her husband, Itsik Pe'er, is a computational biologist at Columbia University. Together, they have raised two daughters.[4]
She subsequently performed postdoctoral work with George Church at Harvard.[8][9][10] Her fellowship focused on how genetic variation changes the regulatory network between individuals and how this subsequently manifests in phenotypic diversity.[11][12][13]
Career
In 2006, Pe'er established a research group in the Department of Biological Sciences and Systems Biology at Columbia University. Pe'er's group at Columbia developed computational methods that combine diverse sources of high throughput genomics data, with the aim of developing a holistic view of the cell at a systems level.[14]
Pe'er is also involved in the Human Cell Atlas as a member of the organizing committee, co-chair of the Analysis Working Group, and member of the Human Lung Cell Atlas initiative, and serves on the scientific advisory board of scverse.[17]
Research
In her PhD work, Pe'er demonstrated that Bayesian networks can describe interactions between thousands of genes, enabling the analysis of data from newly available DNA microarrays, which generate thousands of noisy measurements of gene expression.[18] The approach has been widely applied to genome-scale sequencing data. In her postdoctoral work, she used this framework to study protein signaling networks in multivariate flow cytometry data.[19]
At Columbia, Pe'er applied Bayesian networks to integrate different data types for the study of gene regulatory networks, determining how DNA sequence variation alters the regulation of gene expression, with a view towards personalized medicine.[20]
The Pe'er research group has developed a series of methods for high-throughput single-cell data analysis, initially to address a new high-dimensional data type derived from mass cytometry, which quantifies a few dozen proteins per cell for millions of cells at a time. They introduced the application of non-linear dimensionality reduction by t-distributed stochastic neighbor embedding (t-SNE) to visualize high-dimensional single-cell RNA sequencing data,[21] and the use of a nearest neighbors graph to represent the data manifold of RNA-defined cell states.[22] The Pe'er group used this formalization to identify discrete cell types or cell states by applying the Louvain community detection method to cluster data,[23] and demonstrated that cells can be ordered along differentiation trajectories from individual samples, due to the asynchrony of cells found in tissue samples.[22] By modeling trajectories as a Markov process, they showed that cells can be assigned probabilities for reaching any given terminal fate along a trajectory.[24] In 2020, the Pe'er and Fabian Theis groups presented CellRank, an algorithm that uncovers cellular dynamics by combining trajectories based on cell-cell similarity with local RNA velocity information, which identifies nascent transcriptional states by the proportion of spliced-to-unspliced RNA transcripts.[25]
Pe'er applies these methods to model biological questions around cellular plasticity and single-cell phenotypic variation in cancer, developmental biology, and immunology, including tumor microenvironments,[26]metastasis[26] and responses to treatments such as immunotherapy. "We are beginning to understand that plasticity is a key hallmark of cancer," said Dr. Pe'er. "It is the cancer cell's plasticity that allows it to make such a switch to survive."
Upon accepting the International Society for Computational Biology's Overton Prize in 2014, Pe'er said, "Math is rigorous, and biology is messy, so the trick is to find the pattern in the mess, and machine learning provides a powerful toolbox."[13]
Selected publications
Friedman, Nir; Linial, Michal; Nachman, Iftach; Pe'Er, Dana (2000). "Using Bayesian Networks to Analyze Expression Data". Journal of Computational Biology. 7 (3–4): 601–620. doi:10.1089/106652700750050961. PMID11108481.
